The heat rolling laminar cooling process has the complex natures, such as the highly nonlinearity, the difficulty of the online measurement of the strip temperature continuously in the cooling process and the variation of the heat transfer parameters due to the changing of the operating conditions. For the discrete dynamic model of the strip temperature during the laminar cooling process, the correct identification of the model coefficients is the key factor to the precision of this model. A hybrid intelligent identification algorithm is developed by combining the RBF neural networks, CBR and fuzzy logic reasoning. The tests using real industrial data of a steel plant have been conducted where the results indicate that the proposed hybrid intelligent parameter identification approach has made a great contribution in improving the prediction precision of the strip temperature during the laminar cooling process.
[1]
L. Biegler,et al.
Simultaneous solution and optimization strategies for parameter estimation of differential-algebraic equation systems
,
1991
.
[2]
Sankar K. Pal,et al.
Unsupervised feature evaluation: a neuro-fuzzy approach
,
2000,
IEEE Trans. Neural Networks Learn. Syst..
[3]
Ian D. Watson,et al.
An Introduction to Case-Based Reasoning
,
1995,
UK Workshop on Case-Based Reasoning.
[4]
Chai Tianyou,et al.
INTELLIGENT OPTIMIZATION CONTROL FOR LAMINAR COOLING
,
2002
.
[5]
R. Fletcher,et al.
Hybrid Methods for Nonlinear Least Squares
,
1987
.